Skip to content

docs: add multi-GPU guide and surface Harvey LAB showcase - #468

Open
ShubyM wants to merge 1 commit into
mainfrom
multi_gpu_docs_review
Open

ShubyM wants to merge 1 commit into
mainfrom
multi_gpu_docs_review

Conversation

@ShubyM

@ShubyM ShubyM commented Nov 19, 2026

Copy link
Copy Markdown
Collaborator

Adds a concise multi-GPU guide and wires the 8-GPU Harvey LAB showcase into the main docs:

  • docs/multi-gpu.md (new): documents openrl.trainer_backend=automodel, openrl.trainer_device_count (and deprecated openrl.trainer_gpus alias), openrl.trainer_cp, openrl.sampler_replicas, openrl.exclusive, and openrl.recipe=rl, common layouts, the 8-GPU Harvey LAB showcase, enforced restrictions/exclusivity rules, and placement/utilization checks.
  • docs/setup/gke-setup.md: adds a collapsible 8×H100 (a3-highgpu-8g) DRA node-pool snippet for multi-GPU / Harvey LAB clusters.
  • docs/setup/harvey-lab-on-gke.md & docs/configuration.md: documents OPEN_RL_AUTOMODEL_IMAGE (ghcr.io/openrl-dev/openrl/automodel:latest) and cross-links docs/multi-gpu.md.
  • README.md & docs/getting-started.md: links the Harvey LAB RL recipe, docs/multi-gpu.md, docs/configuration.md, and docs/setup/harvey-lab-on-gke.md.

Verified make lint and all local markdown links.

Signed-off-by: Shuby Mishra <shuby@google.com>

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant